Author: jstephens
Date: 2007-12-06 18:27:59 +0100 (Thu, 06 Dec 2007)
New Revision: 9158
Log:
COUNT() is not consistent on Cluster replication slaves (Thanks, Jonas!)
Modified:
trunk/refman-5.1/mysql-cluster-limitations.xml
trunk/refman-6.0/mysql-cluster-limitations.xml
Modified: trunk/refman-5.1/mysql-cluster-limitations.xml
===================================================================
--- trunk/refman-5.1/mysql-cluster-limitations.xml 2007-12-06 17:25:05 UTC (rev 9157)
+++ trunk/refman-5.1/mysql-cluster-limitations.xml 2007-12-06 17:27:59 UTC (rev 9158)
Changed blocks: 1, Lines Added: 25, Lines Deleted: 0; 1493 bytes
@@ -744,6 +744,31 @@
</formalpara>
</listitem>
+ <listitem>
+ <formalpara>
+
+ <title>Transactions and the <literal>COUNT()</literal> function</title>
+
+ <para>
+ When using MySQL Cluster Replication, it is not possible
+ to guarantee the transactional consistency of the
+ <literal>COUNT()</literal> function on the slave. In other
+ words, when performing on the master a series of
+ statements (<literal>INSERT</literal>,
+ <literal>DELETE</literal>, or both) that changes the
+ number of rows in a table within a single transaction,
+ executing <literal>SELECT COUNT(*) FROM
+ <replaceable>table</replaceable></literal> queries on the
+ slave may yield intermediate results. This is due to the
+ fact that <literal>SELECT COUNT(...)</literal> may perform
+ dirty reads, and is not a bug in the
+ <literal>NDB</literal> storage engine. (See Bug #31321 for
+ more information.)
+ </para>
+
+ </formalpara>
+ </listitem>
+
</itemizedlist>
</para>
Modified: trunk/refman-6.0/mysql-cluster-limitations.xml
===================================================================
--- trunk/refman-6.0/mysql-cluster-limitations.xml 2007-12-06 17:25:05 UTC (rev 9157)
+++ trunk/refman-6.0/mysql-cluster-limitations.xml 2007-12-06 17:27:59 UTC (rev 9158)
Changed blocks: 1, Lines Added: 25, Lines Deleted: 0; 1493 bytes
@@ -726,6 +726,31 @@
</formalpara>
</listitem>
+ <listitem>
+ <formalpara>
+
+ <title>Transactions and the <literal>COUNT()</literal> function</title>
+
+ <para>
+ When using MySQL Cluster Replication, it is not possible
+ to guarantee the transactional consistency of the
+ <literal>COUNT()</literal> function on the slave. In other
+ words, when performing on the master a series of
+ statements (<literal>INSERT</literal>,
+ <literal>DELETE</literal>, or both) that changes the
+ number of rows in a table within a single transaction,
+ executing <literal>SELECT COUNT(*) FROM
+ <replaceable>table</replaceable></literal> queries on the
+ slave may yield intermediate results. This is due to the
+ fact that <literal>SELECT COUNT(...)</literal> may perform
+ dirty reads, and is not a bug in the
+ <literal>NDB</literal> storage engine. (See Bug #31321 for
+ more information.)
+ </para>
+
+ </formalpara>
+ </listitem>
+
</itemizedlist>
</para>
Thread |
---|
• svn commit - mysqldoc@docsrva: r9158 - in trunk: refman-5.1 refman-6.0 | jon | 6 Dec |